Denmark

[DK] First extended collective licence approval for text and data mining and the use of works by AI

IRIS 2026-3:1/5

Terese Foged

Legal expert

On 8 December 2025, the Danish Ministry of Culture granted its first approval of an extended collective licence for a collective management organisation concerning text and data mining and the use of works in AI tools.

Under section 11 b of the Danish Copyright Act, which corresponds to Article 4 of the Digital Service Market (DSM) Directive, anyone with lawful access to a work may extract and reproduce copies of the work for the purposes of text and data mining, provided that the rights holder has not expressly reserved the use of the work in an appropriate manner.

Several key rights holder organisations, including Koda and Danske Forlag (Danish Publishers), have recommended that their members reserve their rights (opt out) in relation to text and data mining. Major players such as DR (the Danish Broadcasting Corporation) and JP/Politikens Hus (newspapers) have also reserved their rights. Against this background, a significant number of Danish rights holders have, in practice, made use of the opt-out option in section 11 b.

Once rights holders have opted out of the text and data mining copyright exception, they may still license text and data mining uses of their works. Such licences may be individual or collective, and collective agreements may be accompanied by an extended collective licence, which entails a more far-reaching effect.

The 8 December 2025 approval of the collective management organisation Tekst & Node is the first approval of extended collective agreements concerning text and data mining and AI, or more generally the intersection between copyright and AI. This extended collective licence marks the beginning of the commercialisation of copyright in relation to AI.

The extended collective licence and the agreements offered by Tekst & Node are limited to the use of literary works in AI tools for internal use, specifically “for teaching purposes in primary schools or internal administrative use in educational institutions” and “internal use in, among other things, institutions and companies’ AI tools”. Accordingly, the extended collective licence and the agreements offered by Tekst & Node must be assumed to cover situations where the text and data mining copyright exception does not fully apply, in addition to covering text and data mining opt-outs.

As follows from Article 12(3)(c) of the DSM Directive and the Danish Copyright Act, rights holders may opt out of this licensing mechanism. If so, this would constitute a double opt-out, namely an opt-out of the text and data mining copyright exception followed by an opt-out of the extended collective licensing mechanism.

However, there is a significant difference in the economic and control implications of not exercising the two types of opt-out. If opt-out is omitted in the text and data mining regime, users can perform text and data mining without having to meet specific conditions, such as reporting and payment of remuneration, because text and data mining falls within an exception to copyright. If opt-out is omitted in the extended collective licence scheme, however, the exploitation takes place under fixed contractual terms and against payment of remuneration, of which the rights holder is entitled to a share, because this is considered rights management under copyright law.